EP2975643A1 - Low-noise cdhgte photodiode array - Google Patents

Abstract

The invention relates to a planar photodiode array (200) comprising a useful layer (10) of Cd x Hg 1-x Te. According to the invention: the useful layer (10) comprises at least two superimposed doped layers (10a, 10b), each interface between two doped layers forming a single PN junction (10ab); the useful layer (10) has at least one separation region (14), extending from the upper face of the useful layer, and separating at least two useful volumes through the PN junction; and beyond a predetermined depth in the useful layer, the average cadmium concentration in the useful volumes is lower than the average cadmium concentration in the separation region (14). The invention also relates to a method of manufacturing such a matrix of photodiodes (200).

Description

TECHNICAL AREA

The present invention relates to the field of CdHgTe infrared photodiodes used to detect infrared radiation.

STATE OF THE PRIOR ART

Various types of photodiode arrays are known in the prior art, in particular photodiode arrays made in a layer of semiconductor material made of a cadmium, mercury and tellurium alloy Cd _x Hg _1-x Te, with x a real between 0 and 1, the terminals being excluded. Throughout the text, this layer of semiconductor material is called "useful layer".

These photodiodes are used for example in the spatial field, for the detection of infrared radiation, in particular the Middle Infrared (MWIR), corresponding to wavelengths of between 4 μm and 5 μm. at 80 K) and the far infrared (or LWIR for English "Long Wave InfraRed", corresponding to wavelengths greater than 8 microns to 80 K).

As a variant, these photodiodes can also be used for the detection of infrared radiation known as SWIR (for "Small Wave Infrared", corresponding to wavelengths between 2 and 3 μm at 80 K), and infrared radiation known as VLWIR (for "Very Long Wave Infrared", corresponding to wavelengths greater than 14 μm to 80 K).

For example, rectangular matrices comprising 640 × 5 12 photodiodes are produced for a step 15 μm (width of a photodiode).

A photodiode generally has a so-called dark current. The dark current is the residual electric current of a photodetector in the absence of illuminance.

A mesa-structure photodiode is a non-planar photodiode, that is to say having topological variations of the upper side of the useful layer, on a so-called useful face. Each emerging part is called "mesa" between the trenches. The technological realization of a mesa structure photodiode therefore requires the creation of trenches in the useful layer. In particular, a matrix of photodiodes mesa structure is made from a stack of two doped layers, the interface between the two doped layers forming a PN junction. Trenches, called "grooves", are then etched in the stack of doped layers. This etching makes it possible to separate pads each having a PN junction and each corresponding to a photodiode. In this type of technology, the etching of the trenches can induce the presence of defects increasing the current of darkness. In particular, the etching of the stack of doped layers creates material defects at the etched interfaces. These material defects lead to spontaneous creations of electron pairs holes. An electron-hole pair corresponds to the appearance in the useful layer of a minority carrier (the electron or the hole). When the minority carrier crosses the PN junction, an electric current is measured which does not correspond to the absorption of electromagnetic radiation. For this reason we speak of a current of darkness. An increase in the dark current prevents, for example, the detection of very low infrared radiation.

An object of the present invention is to provide a matrix of photodiodes Cd _x Hg _1-x Te made from a stack of doped layers, and having a reduced dark current.

Another object of the present invention is to provide a method of manufacturing such a matrix of photodiodes.

STATEMENT OF THE INVENTION

This objective is achieved with a matrix of photodiodes comprising a useful layer made of a cadmium, mercury and tellurium alloy of Cd _× Hg _{1 -x} Te type, the useful layer having a lower face and an upper face of the opposite side to the lower face.

• la couche utile comprend au moins deux couches dopées superposées, chaque interface entre deux couches dopées adjacentes formant une jonction PN ;
• la couche utile présente au moins une région dite de séparation, s'étendant depuis la face supérieure de la couche utile vers sa face inférieure en traversant ladite jonction PN, la région de séparation séparant au moins deux volumes dits utiles qui s'étendent dans la couche utile aussi profondément que la région de séparation ; et
• au-delà d'une profondeur prédéterminée dans la couche utile, la concentration moyenne en cadmium dans la région de séparation est supérieure à la concentration moyenne en cadmium dans les volumes utiles.

According to the invention, the photodiode array has the following characteristics:

• the useful layer comprises at least two superimposed doped layers, each interface between two adjacent doped layers forming a PN junction;
• the useful layer has at least one so-called separation region, extending from the upper face of the useful layer towards its lower face through said PN junction, the separation region separating at least two so-called useful volumes which extend into the useful layer as deep as the separation region; and
• beyond a predetermined depth in the useful layer, the average cadmium concentration in the separation region is greater than the average cadmium concentration in the useful volumes.

The matrix of photodiodes thus forms a planar matrix, in which a PN junction extends over the entire extent of the useful layer.

This PN junction is formed in one piece. In other words, each interface between two superimposed doped layers forms a single PN junction. This single PN junction is traversed by the separation regions which draw several photodiodes, from this single PN junction. Thus, at least two photodiodes, or pixels of the matrix, share the same PN junction.

Advantageously, the average cadmium concentration in the separation region is greater than the average cadmium concentration in the remainder of the useful layer.

The useful layer may consist of two doped layers each having a doping of different nature.

As a variant, the useful layer may consist of three doped layers forming together two PN junctions, two doped layers having a doping of the same nature surrounding a median doped layer having a doping of different nature, and the separation region crossing the two PN junctions.

Preferably, the separation region has a cadmium gradient decreasing from the upper face of the useful layer and towards its lower face.

The separation region may be separated from the underside of the useful layer by at least a portion of said useful layer.

The useful volumes are advantageously distributed in the useful layer in a regular mesh.

In particular, the useful volumes can be distributed in the useful layer in a square mesh, and separated from each other by a single separation region.

According to an advantageous embodiment, the matrix of photodiodes according to the invention comprises at least one over-doped zone located in a region formed by the intersection between a separation region and the doped layer located on the side of the upper face of the the useful layer, called the upper doped layer, the over-doped zone having a doping of the opposite type to that of said upper doped layer.

• réalisation d'une couche utile en un alliage semi-conducteur de cadmium, mercure et tellure de type Cd_xHg_1-xTe, comprenant au moins une jonction PN située entre deux couches dopées superposées de la couche utile ;
• réalisation, sur la face supérieure de la couche utile, d'une couche dite couche structurée présentant au moins une ouverture traversante, et ayant une concentration en cadmium supérieure à la concentration moyenne en cadmium de la couche utile ;
• recuit de la couche utile recouverte de la couche structurée, réalisant une diffusion des atomes de cadmium de la couche structurée, depuis la couche structurée vers la couche utile, formant ainsi l'au moins une région de séparation.

The invention also relates to a method of manufacturing such a matrix of photodiodes. The method according to the invention comprises the following steps:

• producing a useful layer of a semiconductor alloy of cadmium, mercury and tellurium-type Cd _x Hg _1-x Te, comprising at least one PN junction between two superimposed doped layers of the active layer;
• producing, on the upper face of the useful layer, a so-called structured layer having at least one through opening, and having a cadmium concentration greater than the average cadmium concentration of the useful layer;
• annealing the useful layer covered with the structured layer, diffusing the cadmium atoms of the structured layer, from the structured layer to the useful layer, thereby forming the at least one separation region.

Preferably, the through openings are distributed in the structured layer in a regular mesh.

Said steps of producing a structured and annealed layer advantageously form a production cycle, and at least two manufacturing cycles are implemented.

Annealing can be carried out at a temperature between 100 ° C and 500 ° C.

The annealing can be carried out for a period of between 1h and 100h.

The method according to the invention may comprise a doping step, so as to produce at least one over-doped zone located in a region formed by the intersection between a separation region and the doped layer situated on the upper face side. of the useful layer, said upper doped layer, the over-doped zone having a doping of the opposite type to that of said upper doped layer.

BRIEF DESCRIPTION OF THE DRAWINGS

• la figure 1 illustre de manière schématique un premier mode de réalisation de procédé selon l'invention ;
• la figure 2 illustre selon une vue en perspective un premier mode de réalisation d'une matrice de photodiodes selon l'invention ;
• la figure 3 illustre de manière schématique un deuxième mode de réalisation de procédé selon l'invention ;
• la figure 4 illustre selon une vue en perspective un deuxième mode de réalisation d'une matrice de photodiodes selon l'invention ; et
• la figure 5 illustre selon une vue en coupe un troisième mode de réalisation d'une matrice de photodiodes selon l'invention.

The present invention will be better understood on reading the description of exemplary embodiments given purely by way of indication and in no way limiting, with reference to the appended drawings in which:

• the figure 1 schematically illustrates a first embodiment of the method according to the invention;
• the figure 2 illustrates in perspective view a first embodiment of a matrix of photodiodes according to the invention;
• the figure 3 schematically illustrates a second embodiment of the method according to the invention;
• the figure 4 illustrates in perspective view a second embodiment of a photodiode array according to the invention; and
• the figure 5 illustrates in sectional view a third embodiment of a photodiode array according to the invention.

DETAILED PRESENTATION OF PARTICULAR EMBODIMENTS

We have illustrated in figure 1 and schematically, the steps of a method according to the invention for manufacturing a matrix of photodiodes. The views of the figure 1 are sectional views.

The process of figure 1 uses a useful layer 10 made of a semiconductor material made of an alloy of cadmium, mercury and tellurium Cd _x Hg _1-x Te, with x a real between 0 and 1, the terminals being excluded. In particular, x is less than 0.5. Typically x is between 0.2 and 0.4. In the example of the figure 1 the cadmium concentration in the useful layer 10 is constant in space.

For example, we will try to detect a radiation in the mean infrared said MWIR, which corresponds to a cadmium concentration defined by x = 0.3. In a variant, it is sought to detect infrared radiation far away, says LWIR, which corresponds to a cadmium concentration defined by x = 0.22.

One could also want to detect wavelengths in SWIR or VLWIR.

This useful layer 10 is made on a substrate 11, for example a substrate made of an alloy of cadmium, zinc, tellurium. Such an alloy provides a very advantageous mesh arrangement of the useful layer 10 material with the substrate 11. The substrate 11 is transparent to the wavelengths that it is desired to detect. According to a variant not shown, the substrate is separated from the useful layer before, during or after the implementation of the method according to the invention. Thus, the matrix according to the invention does not necessarily include a substrate.

The useful layer 10 typically has a thickness of between 1 micron and 20 microns, for example 15 microns. The substrate 11 has a thickness of the order of 500 microns. For the sake of clarity of the figures, the thickness of the substrate 11 is undersized in the figures.

The useful layer has for example a parallelepiped shape, in particular a rectangular parallelepiped. In this useful layer is defined an axis z starting from the face 108, connecting the faces 108 and 109, and defining the axis of the depth. The face 109 is a lower face (here in contact with the substrate 11). The face 108 is an upper face, the opposite side to the lower face 109.

Step 100:

During a first step 100, a useful layer 10 is produced, having a PN junction 10ab which extends in a plane, over the entire extent of the useful layer 10. It is in particular a plane orthogonal to the z axis.

The PN junction 10ab is formed here by the interface between the two superimposed doped layers 10a and 10b, each having N or P doping of a different nature. For example, the doped layer 10a is P-doped, and the doped layer 10b is N-doped. The doped layers 10a and 10b together form the layer useful 10, and therefore both extend over the entire extent of the useful layer. The layer 10a may be called the upper doped layer. The layer 10b may be called the lower doped layer.

Throughout the text, a layer is said to be P-doped when it has an excess of "holes", or in other words a defect of electrons (minority carriers). A layer is said to be N doped when it has an excess of electrons, or in other words a defect of holes (minority carriers). Throughout the text, the nature of the doping of a doped layer designates the nature of the majority carriers in said layer. We also speak of a type of doping (N or P).

In the example shown in figure 1 the doped layers 10a and 10b are formed during the growth of the useful layer 10.

For example, a first layer is first deposited on the substrate which contains N doping atoms such as indium atoms. This second layer is then deposited on a second layer which contains dopant atoms P such as arsenic atoms. These two layers form the useful layer 10. An annealing is then carried out at about 400 ° C. to activate the P dopants. This annealing creates mercury gaps throughout the useful layer 10. A heavily doped P layer is thus obtained (thanks to arsenic atoms), over a lightly doped P layer (because of mercury gaps). An annealing is then carried out at about 220 ° C. under saturating mercury pressure to fill the mercury gaps. A highly doped P-doped layer 10a is thus obtained over an N-doped layer 10b.

For all the necessary anneals, the skilled person will be able to carry out annealing under conditions of pressure and temperature which limit the degradation of the material.

In a variant, the doped layer 10a is N-doped with indium atoms during growth, and the doped layer 10b is P-doped. For this, a substrate is deposited on the substrate, for example, which contains doping atoms. arsenic. A second layer is then deposited on the first layer which contains N doping atoms such as indium doping atoms. An annealing is then carried out at about 400 ° C. to activate the P dopants. This annealing creates mercury gaps throughout the useful layer 10. Thus, a strongly doped P layer (thanks to the arsenic atoms) is obtained below a weakly doped layer P (because of the mercury gaps). An annealing is then carried out at about 220 ° C. under saturating mercury pressure to fill the mercury gaps. An N-doped layer 10a is thus obtained over a P-doped layer 10b.

It is also possible to perform the P-doping of the layer 10b by its intrinsic impurities such as mercury gaps. For this, the concentration of mercury vacancies must be controlled by suitable annealing and this annealing must not impact the doping of the N layer.

The doping density (indium atoms) in the N layer is for example of 1.10 ¹⁶ atoms / cm ³ and the doping density (arsenic atoms) in the layer P is equal to or greater than 10 ¹⁸ atoms / cm ³ .

The layer deposits then forming the doped layers 10a and 10b are advantageously produced by a technique known as molecular beam epitaxy. Molecular jet epitaxy is a technique for growing a crystal in which the elements to be deposited on a support are evaporated and will then be deposited on this support. Alternatively, a liquid phase epitaxy technique may be used. Liquid phase epitaxy is a crystal growth technique in which the support is brought into contact with a liquid phase of a desired element which crystallizes on the support. Any other technique for depositing a crystalline layer on a support, for example a chemical vapor deposition, may also be envisaged.

Step 101:

During step 101, a so-called reservoir layer 12 is deposited on the upper face 108 of the useful layer 10, one of the elements of which is cadmium and has a cadmium concentration greater than the average cadmium concentration of the useful layer 10.

The reservoir layer 12 is made of a binary, ternary, or quaternary material, or more. This material advantageously comprises elements belonging to columns II and VI of the periodic table of the elements. This is for example and without limitation CdS, CdSe, CdTe, CdZnSe, CdMnSSe, etc.

The reservoir layer 12 has a thickness of the order of one micrometer, for example between 0.1 μm and 10 μm, in particular 1 μm.

The deposition of the reservoir layer 12 is performed by any known thin film deposition technique.

The reservoir layer 12 thus forms a uniform layer which covers the entire upper face 108 of the useful layer 10.

Step 102:

The reservoir layer 12 is then etched so as to form through openings 120. This step is called structuring, or texturing of the reservoir layer 12. The reservoir layer after structuring forms a so-called structured layer 121. The structured layer 121 therefore has the same concentration of cadmium as the reservoir layer 12. is called "through opening" an opening through a layer from one side to the other, in the direction of the thickness.

The production of the structured layer 121 preferably has a step of physical or chemical etching.

Step 102 is broken down for example into two steps 102a and 102b.

Step 102a:

In a first step 102a, a layer of resin is deposited on the reservoir layer 12, and then the openings are etched in this resin. 130. It is preferably an etching by photolithography. A resin mask 13 is thus formed on the reservoir layer 12.

Step 102b:

In a second step 102b, the reservoir layer 12 is etched through the resin mask 13. Thus, the reservoir layer 12 is etched only at the locations not covered by the resin.

The etching is advantageously a chemical etching, typically a chemical etching using a Bromine solution. The etching depth is adjusted by adjusting a duration of the chemical etching. In practice, the reservoir layer has a different appearance of the useful layer, so we can optically identify the moment from which we can stop the chemical etching.

Any other method of structuring is also conceivable, such as a selective chemical etching between the reservoir layer 12 rich in cadmium, and the useful layer of cadmium alloy, mercury, tellurium. Selective non-chemical or non-chemical non-chemical etching is also possible provided that it does not create additional detrimental defects in the useful layer.

• dépôt 101, sur la face supérieure 108 de la couche utile, d'une couche de réservoir 12 présentant une concentration en cadmium supérieure à la concentration moyenne en cadmium de la couche utile 10 ; et
• gravure 102 d'au moins une ouverture traversante 120 dans la couche de réservoir, formant ainsi une couche dite structurée 121 ;

forment ensemble une étape de réalisation, sur ladite face supérieure 108, d'une couche structurée 121 présentant au moins une ouverture traversante 120, et une concentration en cadmium supérieure à la concentration moyenne en cadmium dans la couche utile 10.The following steps:

• deposit 101, on the upper face 108 of the useful layer, a reservoir layer 12 having a cadmium concentration greater than the average cadmium concentration of the useful layer 10; and
• etching 102 of at least one through aperture 120 in the reservoir layer, thereby forming a so-called structured layer 121;

together form a step of producing, on said upper face 108, a structured layer 121 having at least one through opening 120, and a cadmium concentration greater than the average cadmium concentration in the useful layer 10.

It is also conceivable to make the structured layer 121 by a lift-off technology. This involves, for example, depositing a structured layer of resin on the upper face 108, and covering the whole by the reservoir layer 12. The reservoir layer is therefore deposited on the resin, where the resin is present, and on the useful layer 10 at the through openings in the resin. By removing the resin, the structured layer 121 is obtained.

Step 103:

An annealing is then carried out adapted to the assembly formed by the useful layer 10 and the structured layer 121. This annealing will, for example, be carried out at a temperature of between 100 ° C. and 500 ° C., preferably between 300 ° C. and 500 ° C. C, and for a duration ranging from a few minutes to several hours, for example between 1h and 100h, for example between 1h and 40h. In the example shown in figure 1 , the annealing corresponds to heating at 430 ° for 50h.

The layer corresponding to the structured layer 121 after annealing is called the remaining layer 18. It may be provided to remove the remaining layer 18. For example, a planarization method may be used.

During this annealing, the cadmium atoms of the structured layer 121 will diffuse to the useful layer 10. We can therefore speak of a diffusion annealing. This annealing retains the quality of the crystalline structure of the useful layer.

Thus, under the solid portions of the structured layer 121, separation regions 14 each have a descending cadmium concentration gradient from the upper face 108 to the lower face 109 of the useful layer. The solid portions of the structured layer 121 designate the portions surrounding the through apertures, that is to say possibly the reservoir layer portions 12 remaining after the step 102.

Throughout the text, a constant cadmium concentration in space does not define a concentration gradient.

The concentration gradient is in particular a continuous gradient, without breaking a sudden slope.

The separation regions 14 separate at least two so-called useful volumes 16, which extend into the useful layer 10 as deeply as said separation regions 14. For reasons of readability of the figure 1 , the useful volumes are represented by dashed surfaces.

The separation regions 14 extend into the useful layer 10 from the upper face 108 to the interior of the lower doped layer 10b, that is to say up to the inside of the doped layer comprising the underside of the useful layer. If necessary, this doped layer may be defined as being the doped layer closest to the substrate 11. The separation regions 14 thus cross the PN junction 10ab, where appropriate the junction 10ab closest to the lower face 109 of the useful layer. These separation regions 14 extend for example over at least one third of the thickness of the lower doped layer 10b, preferably at least half. Preferably, each separation region 14 is separated from the underside 109 of the useful layer by a portion 15 of the useful layer. It is also possible for at least one separation region 14 to pass through all of the two doped layers 10a and 10b. In this context, the modulation transfer function (FTM) of the diodes is optimized, in addition to the advantages which will be detailed hereinafter.

Since cadmium atoms have diffused into the separation regions 14, the average cadmium concentration in these regions is greater than the average cadmium concentration in at least one adjacent working volume.

For the same reason, the average cadmium concentration in these regions 14 is greater than the average cadmium concentration in the remainder of the useful layer. In particular, the average cadmium concentration in the regions 14 is greater than the average cadmium concentration in a region of the useful layer intended to form an absorption region of the useful layer, in which the photons at the wavelength that one wishes to detect will form minority carriers.

It will also be possible to retain the criterion according to which, beyond a predetermined depth (here z = 0) in the useful layer 10, the average cadmium concentration in the separation regions 14 is greater than the average cadmium concentration in the useful volumes 16. Without precision, the term "concentration" refers to a concentration by volume.

The following criterion can also be used, considering surfaces in the useful layer, parallel to the upper face 108, and defined by a depth z in the useful layer. Having passed a predetermined depth in the useful layer 10, the intersection of such a surface with the regions 14 has a mean cadmium surface concentration greater than the average cadmium surface concentration of the intersection of this same surface with the rest of the surface. useful layer. In the example shown in figure 1 said predetermined depth is less than the useful layer thickness 10 from the upper face 108 to the PN junction 10ab. In particular, said predetermined depth is defined by z = 0.

The regions 14 all open on the upper face 108.

On the right, graphs 103a and 103b show a concentration of cadmium Cd in the useful layer (abscissa axis) as a function of the depth in the useful layer (ordinate axis).

The graph 103b corresponds to a cross-section in the useful layer along the axis BB 'passing through a useful volume 16. It can clearly be seen that the concentration of cadmium then has a continuous value, and corresponding to the initial cadmium concentration of the useful layer. .

The graph 103a corresponds to a section in the useful layer along the AA 'axis passing through a separation region 14. The cadmium concentration has a gradually decreasing profile from the upper face 108 and toward the lower face 109. For example, this gradient takes a maximum value defined by x = 0.5 in Cd _x Hg _1-x Te, and a minimum value defined by x = 0.22 and corresponding to the initial cadmium concentration of the useful layer. The gradient, and thus the separation region 14, extends into the useful layer up to A. "The point A" is about half the thickness of the doped layer 10b, for example at a distance of 1 μm from the lower face 109.

avec

• n(z, t) la concentration volumique en atomes d'une espèce donnée en fonction de la profondeur z et du temps t ;
• t la durée du recuit ;
• D le coefficient de diffusion de l'atome, avec $\mathrm{D}=D_0{e}^{-\frac{Ea}{kT}},$
  
  T la température du recuit et Ea l'énergie d'activation de la diffusion (quantité d'énergie nécessaire pour lancer le processus de diffusion des atomes).

The diffusion of the atoms, in particular the cadmium atoms, can be approximated by a standard diffusion law of Fick type: $$\mathrm{not}\left(\mathrm{z},,\mathrm{t}\right)=\frac{1}{\sqrt{\mathit{\pi Dt}}}{e}^{-\frac{z^2}{4\mathit{dt}}},$$

with

• n (z, t) the volume concentration in atoms of a given species as a function of depth z and time t;
• t the duration of the annealing;
• D the diffusion coefficient of the atom, with $\mathrm{D}=D_0{e}^{-\frac{E\mathrm{at}}{kT}},$
  
  T the annealing temperature and Ea the diffusion activation energy (amount of energy required to start the atom diffusion process).

It is verified that the cadmium concentration gradient follows a decreasing curve starting from the upper face 108 (point A) and towards the lower face 109 (up to point A "), and having a substantially exponential profile. The separation regions 14 all open on the upper face 108.

Depending on the desired characteristics of the separation region 14, in particular its depth and the profile of the concentration gradient, those skilled in the art will be able to adapt the temperature and the duration of the annealing. It may also play on the shape of the structured layer 121 (size and shape of the through openings, spacing between the through openings), and its thickness.

Several cycles may be provided, each comprising the steps of producing a structured layer 121, and annealing. Between two cycles, we delete advantageously the corresponding layer, after annealing, the structured layer previously formed. Each of the cycles can implement a structured layer of different shape.

In the example shown in figure 1 the PN junction 10ab is made before the annealing step 103.

As a variant, it is possible to provide the PN junction 10ab after the annealing step 103. For example, the upper doped layer 10a is then N-doped with indium atoms. The indium atoms are activated during growth, and doping the layer 10a in an N-type doping.

According to another variant, a single annealing effects both the diffusion of cadmium and the activation of a dopant P. In this case, it is possible, for example, to produce an upper layer 10a comprising growth elements P, which is still not enabled. These P doping elements will be activated at the same time as the cadmium will diffuse from the structured layer to the active layer.

It will then be possible to carry out conventional steps of installation of electrical contact elements in contact with the useful layer. These electrical contact elements make it possible to electrically polarize the photodiodes. For example, a contact is provided for each photodiode, electrically connected to the upper doped layer 10a, and a contact common to all the photodiodes electrically connected to the lower doped layer 10b.

It has been observed that in a Cd _x Hg _1-x Te semiconductor material, the bandgap, called "gap", depends on the cadmium concentration. The higher this concentration, the higher the gap. It is thus understood that the separation regions 14 form with the useful volumes 16 a non-continuous 3D heterostructure.

The difference in energy levels in a separation region 14 and in a useful volume 16 forms a potential barrier. Thus, each separation region 14, thanks to a high concentration of cadmium, forms a potential barrier in depth between two adjacent useful volumes 16.

This potential barrier encloses minority carriers in a useful volume 16 of the useful layer. Each of the useful volumes 16 comprises a portion of the PN junction 10ab between the doped layers 10a and 10b, so that each useful volume 16 corresponds to a photodiode.

According to the prior art, starting from a stack of doped layers and wishing to make a photodiode array, pads each comprising a PN junction were separated by trench etching in the stack. The idea underlying the invention is to separate pads each comprising a PN junction by placing potential barriers between these pads. This avoids any etching step to separate pads. This eliminates the disadvantages associated with engraving. Moreover, it remains on a so-called planar technology, that is to say such that the useful layer has a substantially planar upper face. In other words, the useful layer does not have a strong variation of the topography of its upper face. Planar technology simplifies subsequent steps, called "component packaging" (eg installation of electrical contact elements). We can talk about pixelation of the useful layer 10.

As specified in the introduction, the separation etching of the pads is at the origin of a dark current in the photodiode, whose invention thus makes it possible to overcome.

• des porteurs minoritaires créés dans la couche utile en réaction à l'absorption d'un photon, peuvent avoir tendance à se recombiner au niveau des défauts de matériau de l'interface gravée, et non après traversée d'une jonction PN. Dans ce cas, un photon incident dans la couche utile, et à la longueur d'onde de détection, provoquera bien l'apparition d'un porteur minoritaire, mais pas celle d'un courant mesurable dans une photodiode mésa. Les propriétés optiques de la matrice de photodiodes mésa sont donc dégradées ;
• inversement, la présence de défauts liés à la gravure du mésa peut induire la création de porteurs minoritaires. Par exemple, la présence de défauts dans la zone de charge d'espace (ZCE) d'une photodiode va entraîner la création de porteurs minoritaires par un phénomène de génération-recombinaison dans la ZCE. Ces porteurs vont être à l'origine d'un courant électrique en l'absence de flux photonique. Ces défauts vont donc augmenter le courant d'obscurité de la photodiode et ainsi dégrader les performances de la matrice de photodiodes. D'autres phénomènes que la génération-recombinaison, liés à ces défauts, peuvent être à l'origine d'une augmentation du courant d'obscurité.

The invention also makes it possible to overcome the disadvantages inherent to the mesa geometry, and in particular the other defects related to the etching of the mesa structures:

• minority carriers created in the useful layer in response to the absorption of a photon, may tend to recombine at the material defects of the etched interface, and not after crossing a PN junction. In this case, an incident photon in the useful layer, and at the detection wavelength, will cause the appearance of a minority carrier, but not that of a measurable current in a mesa photodiode. The optical properties of the matrix of mesa photodiodes are therefore degraded;
• conversely, the presence of defects related to the etching of the mesa can induce the creation of minority carriers. For example, the presence of defects in the space charge zone (ZCE) of a photodiode will cause the creation of minority carriers by a generation-recombination phenomenon in the ECA. These carriers will be at the origin of an electric current in the absence of photon flux. These defects will therefore increase the dark current of the photodiode and thus degrade the performance of the photodiode array. Other phenomena than the generation-recombination, related to these defects, can be at the origin of an increase of the current of darkness.

The figure 2 is a perspective view of a matrix of photodiodes 200 obtained using the method of figure 1 . The matrix of photodiodes 200 has a planar geometry.

In the example shown in figure 2 a single separation region 14 formed in one piece allows to isolate all useful volumes (not shown in FIG. figure 2 ). The openings 120 in the remaining layer 18, and therefore the useful volumes, are distributed in the useful layer in a regular mesh, including a square mesh. We speak of regular mesh when all the patterns have the same shape and are regularly spaced in space.

The figure 3 illustrates a second embodiment of the method according to the invention. The numerical references of the figure 3 correspond to the numerical references of the figure 1 , the first digit of each number being replaced by a 3.

The figure 3 will only be described for its differences with regard to the figure 1 .

The useful layer 30 made in step 301 consists of a stack of three doped layers 30a, 30b and 30c, together forming the useful layer 30. doped layers 30a and 30c each have N or P doping, different in nature from that of the layer 30b. The interface 30ab between the doped layers 30a and 30b thus forms a first PN junction. The interface 30bc between the doped layers 30b and 30c forms a second PN junction.

The reservoir layer 32 has a cadmium concentration greater than the average cadmium concentration in the useful layer 30, the useful layer 30 being formed by the three doped layers 30a, 30b, 30c.

At the end of the diffusion annealing 303, separation regions 34 are formed in which cadmium atoms have diffused.

According to a first variant of the embodiment represented in figure 3 beyond a predetermined depth in the useful layer 10 corresponding to the depth of the PN junction 30bc, the average cadmium concentration in the separation regions 34 is greater than the average cadmium concentration in the useful volumes 36.

According to a second variant of the embodiment represented in figure 3 the average cadmium concentration in the separation regions 34 is greater than the average cadmium concentration in the useful volumes 36.

The separation regions 34 extend to the inside of the doped layer 30c, crossing the two PN junctions 30ab and 30bc. In particular, the separation regions 34 pass through the PN junction bc, which is the PN junction closest to the lower face 309 of the useful layer 30. The doped layer 30c of the figure 3 corresponds to the doped layer 10b of the figure 1 .

The figure 4 is a perspective view of a matrix of photodiodes 400 obtained using the method of figure 3 . It corresponds to the matrix 200 illustrated in figure 2 except that it has three doped layers instead of two. The upper doped layer 30a has a thickness of the order of 5 m and an N doping having a doping density of the order of 10 ¹⁷ atoms / cm ³ . The median doped layer 30b has a thickness of the order of 5 μm and a P doping having a doping density of the order of 10 ¹⁷ atoms / cm ³ . The lower doped layer 30c has a thickness of the order of 10 μm and an N doping having a doping density of the order of 10 ¹⁹ atoms / cm ³ .

An advantage of such a matrix of doped three-layer photodiodes is that it forms a stack of two elementary matrices of elementary photodiodes positioned head-to-tail and sharing the same median doped layer 30b. There is thus a series of stacks of two photodiodes positioned head to tail, each stack of two photodiodes corresponding to a stack of two PN junctions positioned head to tail.

By polarizing one or the other of the PN junctions 30ab or 30bc, one or the other elementary matrix of photodiodes is used. Advantageously, an upper doped layer 30a has a first concentration of cadmium, which corresponds to the absorption of a first wavelength, and a lower doped layer 30c has a second concentration of cadmium, which corresponds to the absorption of a second wavelength.

• pour chaque empilement de deux photodiodes, d'un élément de contact électrique relié électriquement à la couche dopée supérieure 30a ; et
• d'un élément de contact électrique commun à tous les empilements de deux photodiodes, relié électriquement à la couche dopée inférieure 30c.

The polarization of PN junctions is typically done using:

• for each stack of two photodiodes, an electrical contact element electrically connected to the upper doped layer 30a; and
• an electrical contact element common to all the stacks of two photodiodes, electrically connected to the lower doped layer 30c.

By reverse biasing the PN junction between the upper doped layer 30a and the middle doped layer 30b, incident photons are detected at the first wavelength. By reverse biasing the PN junction between the lower doped layer 30c and the median doped layer 30b, the incident photons are detected at the second wavelength. The same structure makes it possible to detect two different wavelengths, which is why it is called a "matrix of bispectral photodiodes".

Typically, the lower doped layer 30c has a cadmium concentration defined by x = 0.3 (detection in the MWIR), and the upper doped layer 30a has a cadmium concentration defined by x = 0.22 (detection in the LWIR) . The median doped layer 30b may have a cadmium concentration defined by x = 0.7 (corresponding to the detection of a radiation in the low infrared, towards a wavelength of 1.5 m). The order of the stacking of the layers is a function of their gap. The highest gap framed between two layers of smaller gaps is preferably provided.

We have illustrated in figure 5 a third embodiment of a matrix of photodiodes 500 according to the invention. The numerical references of the figure 5 correspond to the numerical references of the figure 1 , the first digit being replaced by a 5.

• suppression de la couche structurée 521 ;
• sur-dopage local de façon à former au moins une zone sur-dopée 550 dans la couche utile, située à la fois à l'intérieur d'au moins une région de séparation 54 et à l'intérieur de la couche dopée supérieure 50a de la couche utile.

The matrix of photodiodes 500 according to the invention is obtained by carrying out the steps of the method as described with reference to FIG. figure 1 followed by the following steps:

• removing the structured layer 521;
• locally over-doping so as to form at least one over-doped zone 550 in the useful layer, located both within at least one separation region 54 and within the upper doped layer 50a of the useful layer.

Each over-doped zone 550 is made to be located only within at least one separation region 54, i.e. in a region of the useful layer where the cadmium atoms have diffused.

The upper doped layer 50a designates one of the superimposed doped layers together forming the useful layer. The upper doped layer 50a more particularly denotes the doped layer comprising the upper face 508 of the useful layer. Each over-doped zone 550 is made to be located only inside this upper doped layer 50a.

Thus, each over-doped zone 550 is made to be located within a region formed by the intersection of the upper doped layer 50a and a separation region 54.

An over-doped zone 550 may extend over an entire region formed by the intersection of the upper doped layer 50a and a separation region 54. Alternatively, an over-doped zone 550 may extend over only a portion of the region formed by the intersection between the upper doped layer 50a and a separation region 54.

Each over-doped zone 550 has a doping of a different nature from that of the doping of the upper doped layer 50a. For example, either an N-doped upper doped layer, the over-doped zone 550 exhibits P-type doping. Inversely, or a P-doped upper doped layer, the over-doped zone 550 exhibits N-type doping. -dopée 550 preferably has a doping level at least ten times higher than that of the upper doped layer 50a.

In the case where the upper doped layer is N-doped, an over-doped zone 550 may be produced by diffusion or implantation of P doping atoms such as arsenic or phosphorus atoms. If necessary, an activation annealing is then carried out.

In the case where the upper doped layer is P-doped, an over-doped zone 550 can be produced by diffusion or implantation of N doping atoms such as boron or indium atoms. If necessary, an activation annealing is then carried out.

A matrix of photodiodes 500 is thus produced, which forms an advantageous variant of the photodiode matrix as illustrated in FIG. figure 2 .

The at least one over-doped zone 550 makes it possible to locally increase the forbidden band value in the useful layer, in the vicinity of the regions of separation. This strengthens the potential barrier formed by the separation regions 54.

The photodiode matrix thus obtained thus has a further dark current. It also has a further improved MTF.

A variant of the second embodiment of a method and a matrix of photodiodes illustrated in FIG. figures 3 and 4 .

The invention is not limited to the examples just described, and we can imagine many variants, without departing from the scope of the present invention. For example, the useful layer may have, before producing a structured layer, a higher concentration of cadmium over a certain thickness on the side of its upper face. The separation regions may have a particular doping.

Claims (15)

A planar photodiode array (200; 400; 500) comprising a useful layer (10; 30) of a Cd _x Hg _1-x Te cadmium, mercury and tellurium alloy, the useful layer having a lower face ( 109; 309) and an upper face (108; 308; 508) on the opposite side to the lower face; characterized in that the useful layer (10; 30) comprises at least two superimposed doped layers (10a, 10b, 30a, 30b, 30c, 50a, 50b), each interface between two adjacent doped layers forming a PN junction (10ab; 30ab, 30bc; 50ab) which extends over the entire extent of the useful layer; the useful layer (10; 30) has at least one so-called separation region (14; 34; 54) extending from the upper face (108; 308; 508) of the useful layer towards its lower face (109; 309) through said PN junction (10ab; 30ab, 30bc; 50ab), the separation region (14; 34; 54) separating at least two so-called useful volumes (16; 36; 56) which extend into the useful layer as deeply as the region of separation; and - above a predetermined depth in the useful layer, the average cadmium concentration in the separation region (14; 34; 54) is greater than the average cadmium concentration in the useful volumes (16; 36; 56) . A photodiode array (200; 400; 500) according to claim 1, characterized in that the average cadmium concentration in the separation region (14; 34; 54) is greater than the average cadmium concentration in the remainder of the layer useful. Photodiode array (200; 500) according to claim 1 or 2, characterized in that the useful layer (10) consists of two doped layers (10a, 10b; 50a, 50b) each having a different kind of doping. Photodiode array (400) according to claim 1 or 2, characterized in that the useful layer (30) consists of three doped layers (30a, 30b, 30c) forming together two PN junctions (30ab; 30bc), two doped layers ( 30a, 30c) having a doping of the same nature surrounding a medial doped layer (30b) having a doping of different nature, and the separation region (34) crossing the two PN junctions (30ab; 30bc). A photodiode array (200; 400; 500) according to any one of claims 1 to 4, characterized in that the separation region (14; 34; 54) has a descending cadmium gradient from the upper surface (108; 308). 508) of the useful layer and towards its lower face (109; 309). A photodiode array (200; 400; 500) according to any one of claims 1 to 5, characterized in that the separation region (14; 34; 54) is separated from the bottom face (109; 309) of the layer useful by at least a portion (15; 35) of said useful layer. Photodiode array (200; 400; 500) according to any one of claims 1 to 6, characterized in that the useful volumes (16; 36; 56) are distributed in the useful layer (10; 30) in a regular mesh. . A photodiode array (200; 400; 500) according to claim 7, characterized in that the useful volumes (16; 36; 56) are distributed in the useful layer (10; 30) in a square grid and separated from each other. others by a single separation region (14; 34; 54). A photodiode array (500) according to any one of claims 1 to 8, characterized in that it comprises at least one over-doped zone (550) located in a region formed by the intersection of a separation region ( 54) and the doped layer located on the side of the upper face of the useful layer, called the upper doped layer (50a), the over-doped zone having a doping of the opposite type to that of said upper doped layer (50a). A method of manufacturing a planar photodiode array (200; 400; 500) according to any one of claims 1 to 9, characterized in that it comprises the following steps: - Making (100; 300) a useful layer (10; 30) of a Cd _x Hg _1-x Te cadmium, mercury and tellurium semiconductor alloy comprising at least one PN junction (10ab; 30ab, 30bc; 50ab), located between two superposed doped layers (10a, 10b; 30a, 30b, 30c; 50a, 50b) of the useful layer and which extends over the entire extent of the useful layer; - realization (101, 102; 301, 302), on the upper face (108; 308; 508) of the useful layer (10; 30), a so-called structured layer layer (121; 321) having at least one opening through (120; 320), and having a cadmium concentration higher than the average cadmium concentration of the useful layer (10; 30); annealing (103; 303) of the useful layer (10; 30) covered with the structured layer (121; 321), diffusing the cadmium atoms of the structured layer (121; 321) from the structured layer (121; 321) to the useful layer (10; 30), thereby forming the at least one separation region (14; 34; 54). Method according to claim 10, characterized in that the through openings (120; 320) are distributed in the structured layer (121; 321) in a regular mesh. A method according to claim 10 or 11, characterized in that said steps (101, 102, 301, 302) of a structured layer (121; 321), and annealing (103; 303) form a production cycle, and in that at least two manufacturing cycles are carried out. Process according to any one of Claims 10 to 12, characterized in that the annealing (103; 303) is carried out at a temperature of between 100 ° C and 500 ° C. Process according to Claim 13, characterized in that the annealing (103; 303) is carried out for a period of between 1 h and 100 h. Process according to any one of Claims 10 to 14, characterized in that it comprises a doping step, so as to produce at least one over-doped zone situated in a region formed by the intersection between a separation region and the doped layer located on the side of the upper face of the useful layer, said upper doped layer, the over-doped zone having a doping of the opposite type to that of said upper doped layer.

Images